add support for new organization-specific workspace limit setting

[SwiftEngineer]

Description

This change allows go-tfe to read/set/unset a newly created setting on the organization model. This setting is only available in TFE.

Testing plan

1. Run local development version of TFE
2. Run the newly updated tests in admin_organization_integration_test.go

Output from tests

It's worth noting that I had to make a few changes to the existing test (specifically, assertions about the GlobalModuleSharing flag) in order to get these tests to pass. Since those tests don't get run in CI, I'm thinking they probably haven't run in a long while.

Looking at #278, I can't figure out how the assertions could have passed, since globalModuleSharing was changed from a bool to *bool, shouldn't the assertions reflect that? Maybe the test library changed? I have no idea! 😅 This is my first PR for Golang, so if anyone could correct me here, I can happily revert my changes there and fix whatever is going on.

go tool test2json -t tmp/path-from/GoLand/___TestAdminOrganizations_Update_fetches_and_updates_organization_in_github_com_hashicorp_go_tfe.test -test.v -test.paniconexit0 -test.run ^\QTestAdminOrganizations_Update\E$/^\Qfetches_and_updates_organization\E$
=== RUN   TestAdminOrganizations_Update
--- PASS: TestAdminOrganizations_Update (3.37s)
=== RUN   TestAdminOrganizations_Update/fetches_and_updates_organization
    --- PASS: TestAdminOrganizations_Update/fetches_and_updates_organization (2.45s)
PASS

[SwiftEngineer]

I added a string version of nullable as well (even though it isn't in use right now) so that folks would have an easier time understanding how they might extend Nullable to work with their own types.

[JarrettSpiker]

LGTM.
Im assuming that 0 is handles separately from null on the backend, which is why we want the value to be either?

My only nitpick with that it is nil and not null in go. I guess it shows up as null in the JSON though, so 🤷

[SwiftEngineer]

LGTM. Im assuming that 0 is handles separately from null on the backend, which is why we want the value to be either?

Can confirm, that's a valid and true assumption 👍

My only nitpick with that it is nil and not null in go. I guess it shows up as null in the JSON though, so 🤷

Tell me about it. I did write a test to verify that at least!

[SwiftEngineer]

small little typo fix that I didn't think was worth a second PR, so I snuck it in here 🥷

[barrettclark]

@SwiftEngineer Would you mind squashing the commits once you're ready for this to be merged in?

[SwiftEngineer]

@SwiftEngineer Would you mind squashing the commits once you're ready for this to be merged in?

No sweat! will do

[brandonc]

Thanks! Be sure to update the changelog and it looks g2g

[brandonc]

I'm a little confused by the argument swapping here and on L241. The other assertions seem to put the parameter second and the model first. And also by the address operator. What's going on?

[brandonc]

It's worth noting that I had to make a few changes to the existing test (specifically, assertions about the GlobalModuleSharing flag) in order to get these tests to pass. Since those tests don't get run in CI, I'm thinking they probably haven't run in a long while.

OK ! Nevermind !

[annawinkler]

Please squash the commits & then this looks great!

[github-actions]

Reminder to the contributor that merged this PR: if your changes have added important functionality or fixed a relevant bug, open a follow-up PR to update CHANGELOG.md with a note on your changes.